Seagate Momentus 5400.4 failure - with youtube video

[Chico]

Hello all.

I have just joined this forum hoping that you'll be able to point me out in the right direction.

I have a Seagate Momentus 5400.4 250Go drive, ST9250827AS, firmware 3.AAA

A couple of days ago when waking up my laptop after hybernation the drive crashed.
The symptoms are that it causes an abnormal delay on the POST, after which the drive is recognized. But then no OS is booted.
I have thus taken the drive out and put it into an external enclosure. The drive spins up immeditly with no alarming sounds or clicks but isn't instantly recognized by windows XP. Then after around 30secs windows XP makes it usual "usb-device-connected-sound" and the drive makes some repeated clicks. After a while a drive pops up in the "My Computer" window, but it shows up as a RAW drive, with no informations whatsoever. It doesn't appear in the Computer Management window either, under "Disk Management" nothing appears. So I booted Ubuntu. Two partitions are now recognized, with the right names, but when I try to mount them it takes a long time and then ubuntu can't mount them and the icons disappear. I can write down the exact error messages on ubuntu, if you like.

So I tried using Spinrite. It recognizes the drive but I try running a test and it stays permantly at 0%.
So I tried using Seatools from Seagate. When doing a basic test, it blocks after 10%, when doing the long one it says that the drive is damaged.

I have uploaded a video of the clicks when plugging the drive on a usb enclosure.

http://www.youtube.com/watch?v=6K_YMDf32iw

Please note that there is a lot of background noise.

The drive spins up at around 0:06 and starts making the repeated clicks after around 0:32. It makes those clicks for a couple dozen seconds more then it stops making them (but never shuts down).

Would anyone be able to recognize the sounds and point me out on the right procedure and help me troubleshooting?

I can try to get a donor PCB if you think it can be a problem with the PCB. I have unscrewed the PCB and could see nothing burned or smelly with it, though.
I can buy one of those cables to run the seagate terminal microcodes if it is a problem that can be solved via this. Could it be a firmware issue?

One of the times I tried plugging it into ubuntu it said something about the NTFS being corrupted and that there was a problem with the file hiberfil.sys.
Bad blocks? Parked heads? What can be the problem?

I can try to get the same drive with same FW and transplant the platters but I need your help in finding out the exact failure, I don't want to open it up without being 100% sure. I have also put it into the freezer for 3h, with no success.

I have a recent backup of this hard drive but ironically enough it's missing some vital data that is very important for my girlfriend's studies. We can't obviously afford to pay for professional services that take 100s of euros/ dollars.
I monitor SMART information often via CrystalDiskInfo and never spotted anything wrong or any bad sectors. The drive never made any strange noises, and I have heard quite a few wrong noises from other drives. It never had any abnormal behaviour. The drive is still under warranty from Seagate until 2013 too.

Looking forward to hearing from you, thanks in advance.

Best regards,

Chico

[poehere]

You should stop now. You have some major damage on this HDD. You best bet if you want to continue on is to image this drive with maybe dd_rescue. Then you can go with a DR recovery program and see if you can find any data on it at all. But you will need to find an adaptor and put this one your PC in order to do this one. I would not try and continue on playing around with this drive in that conditions. Image it ASAP.

[Chico]

Hi there. Thanks for your advice. But how can I image the drive if I can't access it from anywhere? I'll have a look at this dd_rescue.
I know the DD commande from GNU/Linux but I don't understand how I can achieve this if I can't mount the drive?

Anyway, do you have any idea on what the damage might be? This laptop didn't fall or whitstand any shocks, it's mostly used as a desktop.

[Vulcan]

Depending on the exact internal problem (which is impossible to diagnose with certainty from sound or video), you might only have one attempt to get the data off the drive, but if you are determined not to use pro services and to take the risk yourself, then my comments are below...

The drive spins up immeditly with no alarming sounds or clicks

Good - drive likely inits OK.

but isn't instantly recognized by windows XP. Then after around 30secs windows XP makes it usual "usb-device-connected-sound" and the drive makes some repeated clicks. After a while a drive pops up in the "My Computer" window, but it shows up as a RAW drive, with no informations whatsoever.

This suggests problems reading the filesystem metadata.

I can write down the exact error messages on ubuntu, if you like.

The messages would have been helpful, but due to the possibility of the drive degrading with each attempt to mount the filesystems, it's not worth the risk now IMHO. There are clealy read problems with the drive, which are outside of a DIY fix.

So I tried using Spinrite.

Very bad idea - Spinrite tries to write to the drive, and you never do that in a data recovery situation. Thankfully it seems to have been unsuccessful.

So I tried using Seatools from Seagate. When doing a basic test, it blocks after 10%, when doing the long one it says that the drive is damaged.

Yes, this all fits with problems reading from the media.

Could it be a firmware issue?

No.

Bad blocks?

Yes - caused by an unknown internal problem.

Parked heads?

If you understood enough about how disk drives work, to even think about transplanting platters, you would not have asked this

I can try to get the same drive with same FW and transplant the platters

No!!! Transplanting platters is not the usual response, no matter what the internet videos show you. You will likely lose you data permanently if you attempt this.

We can't obviously afford to pay for professional services that take 100s of euros/ dollars.

It's your data, so it's your choice and your risk as to what you do. So far in your posting, you have shown me that you are not ready to do any mechanical work with the disk, as you don't know enough about disk drive functionality. That's fine, that's why you're asking here but seeing internet videos of platter transplants does not mean that anyone should do them, or that they are sensible for this situation.

I monitor SMART information often via CrystalDiskInfo and never spotted anything wrong or any bad sectors.

This is perfectly normal - who told you that you would get a SMART warning before all drive problems?

The drive is still under warranty from Seagate until 2013 too.

Warranty does not cover data recovery or consequential losses (like loss of data).

As you saw already:

You should stop now. You have some major damage on this HDD. You best bet if you want to continue on is to image this drive with maybe dd_rescue. Then you can go with a DR recovery program and see if you can find any data on it at all. But you will need to find an adaptor and put this one your PC in order to do this one. I would not try and continue on playing around with this drive in that conditions. Image it ASAP.

Thanks for your advice. But how can I image the drive if I can't access it from anywhere? I'll have a look at this dd_rescue.
I know the DD commande from GNU/Linux but I don't understand how I can achieve this if I can't mount the drive?

Stop, stay calm, and read about dd_rescue (or the alternative called ddrescue). Google is your friend. You do NOT mount the original disk, before using them! So your problem with mounting the original disk (which is expected) makes no sense, because you are using the wrong procedure It does not matter that you say you know dd - you are not understanding what pohere told you. Stop guessing and/or rushing, as you risk making a mistake and increasing the possibility of losing your data.

As you were told by pohere, the best DIY action you can try as the next step, is to clone the original disk (that means the raw disk e.g. /dev/sda or whatever it is called on your system - do not try to mount it!!!!!!) onto a new blank disk, which is at least the size of the full capacity of the original disk (250GB) using software which does not stop on unreadable sectors (e.g. dd_rescue or ddrescue when using Linux).

There are risks to this procedure - if you look back to some of my previous postings, I have explained them before. Also you must get the direction of copy correct (from your old disk, to your blank disk). There might be other steps you need to take after that, but cloning the original disk onto a new blank disk is the first step.

I am unlikely to have time to assist further on this, but at least I was able to answer your problems about mounting the drive in Linux. Good luck - you'll need it.

[Chico]

Hi Vulcan! Thanks for your input, I appreciate it.

To make it short, I'll just reply to some of your comments.

If you understood enough about how disk drives work, to even think about transplanting platters, you would not have asked this

Ok ok! Sorry for the naif idea, please don't bash me

This is perfectly normal - who told you that you would get a SMART warning before all drive problems?

Nobody did. But apart from monitoring SMART is there any warnings to be monitored anywhere else? (apart from general drive behaviour, noises, etc)

Warranty does not cover data recovery or consequential losses (like loss of data).

I know this. I was just trying to give all the information that ran across my mind. I know, this was not relevant, sorry.
If I get to recover the important data with dd, then i'll be very happy that I didn't rip open the warranty and I can RMA the drive. But that would just be the icing on top of the cake, for the time being I'll just be happy to recover anything

Stop, stay calm, and read about dd_rescue (or the alternative called ddrescue). Google is your friend. You do NOT mount the original disk, before using them! So your problem with mounting the original disk (which is expected) makes no sense, because you are using the wrong procedure

Ok, great! It's one of those cases where I love being wrong

It does not matter that you say you know dd - you are not understanding what pohere told you.

OK, what I meant by "knowing dd" is that I sorta know who dd was, not that we hang out a lot together or anything. I understand what pohere told me, now. But not knowing dd that well and not knowing pohere either, I assumed that maybe pohere's advice was given a bit too fast (since I assumed dd needed to mount the partitions and that wasn't possible in my case).

Stop guessing and/or rushing, as you risk making a mistake and increasing the possibility of losing your data.

I was just trying to ask a question about dd, no matter how awkward that might have appeared to you. I don't know how to formulate perfect questions, sorry again.

As you were told by pohere, the best DIY action you can try as the next step, is to clone the original disk (that means the raw disk e.g. /dev/sda or whatever it is called on your system - do not try to mount it!!!!!!) onto a new blank disk, which is at least the size of the full capacity of the original disk (250GB) using software which does not stop on unreadable sectors (e.g. dd_rescue or ddrescue when using Linux).

So this is what I shall do!
I have been doing some reading on dd_rescue. If you would by any chance happen to have any good links handy, by all means do not hesitate! I have a 1To drive that can receive the copied data. It seems the best thing is to wipe it, rather than have dd create an image file on an existing partition, did I get this right?

There are risks to this procedure - if you look back to some of my previous postings, I have explained them before.

ok, will search for them now. But I guess this is my only shot right?

I am unlikely to have time to assist further on this, but at least I was able to answer your problems about mounting the drive in Linux. Good luck - you'll need it.

No problem! All help is good! Thanks, once again. I feel much more reassured now than some hours ago. HOPE!

[Chico]

you will need to find an adaptor and put this one your PC in order to do this one. I would not try and continue on playing around with this drive in that conditions. Image it ASAP.

Poehere, I am going to try to image the drive with dd. What do you think is the best way to do so? I have an usb adaptor. I can boot the laptop with a bootable linux cd that has dd_rescue and have the momentus drive on the usb adaptor and the receiver drive on the primary sata controller of the laptop, or the other way around, or I can have both of them on usb. Apart from the speed of reading/writing over USB, is there a preferable setup? Just checking as I'm not exactly sure what you meant by "you will need to find an adaptor etc".

Also, do you have any suggestions on how I should prepare the receiving hard disk - format it in fat32, NTFS, ext3, leave it unformatted?

[Chico]

News from the recovery-attempt front

OK so I did some reading on dd_rescue.
Actually, a lot of websites suggested using ddrescue rather than dd_rescue.
I donwloaded ubuntu rescue remix. Installed the faulty drive inside a desktop PC and a 1TB drive on an external esata enclosure (spent all morning backing up the 1TB drive and wiping it clean) formatted in ext3.
Booted the CD.
During boot, first error messages:

ntfs_attr_pread_i: ntfs_pread failed: Input/output error
Failed to read hiberfil.sys: Input/output error
Failed to sync device /dev/sda1: Input/output error
Failed to mount '/dev/sda1': Input/output error
NTFS is either inconsistent, or there is a hardware fault, or it's a
SoftRAID/FakeRAID hardware. In the first case run chdsk /f on Windows
then reboot into Windows twice. The usage of the /f paramter is very
important ! If the device is a SoftRAID/FakeRAID then first activate
it and mount a different device under the /dev/mapper/ directory, (e.g.
/dev/mapper/nvidia_eahaabcc1). Please see the 'dmraid' documentation
for more details.
stdin: I/O error

(These are the exact same messages that popped up when I tried to mount the partitions on a previous ubuntu live CD, under Gnome, so there you go!)

(...)

ata_id[939]: HDIO_GET_IDENTITY failed for '/dev/sda'

ata_id[1021]: HDIO_GET_IDENTITY failed for '/dev/sda'

I finally make it to a bash.
fdisk doesn't want to hear from me (forgot to write down the error), so I made this instead:

ubuntu@ubuntu:~$ sudo lshw -C disk - short

result:

[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] Buffer I/O error on device sda, logical block 1
[661.995442] Buffer I/O error on device sda, logical block 2
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 3
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] Buffer I/O error on device sda, logical block 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 128
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 0
[661.995442] end_request: I/O error, dev sda, sector 0
H/W path Device Class Description
/0/100/e/0 /dev/cdrom disk DVD RW AD-7170S
/0/100/e/0/0 /dev/cdrom disk
/0/100/e/1 /dev/sda disk 250GB SCSI Disk
/0/100/e/0.0.0 /dev/sdb disk 1TB Hitachi HDS72101

(actually it's not [661.995442] on each line, but a different number, would you need me to copy the exact ones?)

So I sent:

ubuntu@ubuntu:~$ sudo ddrescue -r 3 -v /dev/sda /dev/sdb logfile.log

and it produces errors like this ad infinitum until I press ctrl+C

[ 1037.505634] end_request: I/O error, dev sda, sector 522312
etc

any suggestions?

[dick]

any suggestions?

Yea, Check the drive in Mhdd.

[Chico]

Checking it now.

[Chico]

Ok, was scanning it with following results:
<3ms : 71825
<10ms: 13830
<50ms: 146
<150ms: 1

then I'm getting two timeouts:
2832021
2932276

i'm stopping it now.

[Vulcan]

You're asking more questions, more quickly than I have time to respond. Some replies below to your most recent questions - it's all I have time to do at the moment, as the day job is keeping me busy... I'm sure you're missing useful info from earlier questions that I don't have time to answer, since you're asking more & more

Actually, a lot of websites suggested using ddrescue rather than dd_rescue.

Yes indeed - I didn't want to confuse you by changing the recommendation from pohere. Personally, of those two, I prefer ddrescue (which I've modified & recompiled), but you'll find people who prefer dd_rescue too.

During boot, first error messages:
[...]
Failed to mount '/dev/sda1': Input/output error

So your Linux CD is trying to automount /dev/sda - bad, bad, bad Not all live CDs try to mount drives that they find. I suggest you read the docs for your live CD, to see if there is any boot option not to automount devices.

Due to the behaviour you described earlier:

it causes an abnormal delay on the POST, after which the drive is recognized.
[...]
After a while a drive pops up in the "My Computer" window

therefore I expect that the drive is taking a long time to become ready and trying to mount the filesystem by that Linux CD will be unsuccessful until the sick drive has become ready (and mounting it shouldn't be done anyway). So the behaviour you're seeing with Linux is understandable from the initial symptoms you described.

ubuntu@ubuntu:~$ sudo ddrescue -r 3 -v /dev/sda /dev/sdb logfile.log

I guess that means your logfile is going into ramdisk created by the live CD - if so, that's not optimal as it gets lost on a reboot (and hence that loses the point of it's ability to restart ddrescue where it had previously reached).

Also, there was no point in you formatting a partition on /dev/sdb (e.g. /dev/sdb1?) as ext3 which you described, since you have chosen to overwite the raw /dev/sdb with the clone of /dev/sda using the command line above.

Perhaps more sensible would be (after partitioning /dev/sdb again, reformatting /dev/sdb1 & mounting it) would be to clone /dev/sda to a file on that mounted filesystem on the 1TB disk. That would allow more flexibility of how you use the other 750GB on that 1TB /dev/sdb disk during subsequent parts of the recovery process (if you get to those). But that's not relevant to the issue you have of reading from /dev/sda.

IMHO with this drive behaviour, depending on what changes when the drive becomes ready after some time, this may be a real challenge requiring repeated attempts to read the drive at different times - or this may simply be beyond what you can do without the more sophisticated recovery hardware (e.g. Deepspar) which some pros have, or may even require hardware intervention inside the HDA (not recommended for you to attempt).

Once you've stopped attempts by your live CD to mount the filesystems on that drive during boot, your options include:

- attempting a reverse clone; and/or
- preventing all ddrescue retries and "splitting" for at least one pass of an attempted clone; and/or
- using raw access (bypassing the Linux kernel cache) to prevent extra reads (and retries) being done by the kernel.

As you can see, ddrescue (and dd_rescue) are not "1 click" solutions, due to the number of options which they have for how you can use them. Different options are needed for different drive faults & situations.

It's impossible to know in advance what will be successful and what will not, but you have decided to spend lots of time trying this stuff, by deciding to do DIY - so understand that there are no quick or guaranteed solutions If there is media damage, then the more you try to clone the disk yourself, the more damage you'll be doing inside the HDA - we can't know if there is media damage, but you said the data is not worth using a pro for help, so this is all being done at your risk.

I wish you good luck and hope you are successful...

[poehere]

OK if you are having this much trouble with all this it can be that some drives can not be cloned with software alone. You can chose ddrescue or dd_resuce. There are other software tools out there to clone with too on this one.

You might have to have this drive cloned for you by someone who has the right tools to handle it and can get off your data on it for you. If you know of anyone who has a DDI, Atola, DC or DE then any one of these tools can do this job for you. They can handle the bad sectors, they can keep your drive powered on and are designed for drives such as yours in this case. You are in France but I am in French Polynesia. That is way too far away to try and help you out on this one.

If you continue to run into problems on this and can not mount your drive and get it cloned may I make a suggestion as to contact Sean from PCImage he could handle this one for you and if all you are saying on this is true then I feel the price would not be so expensive to get off your data on this drive. He has all the tool necessary to take care of a job like this same as others like me and people in this field. I just do not want to see you kill this drive now after all you have tried to do on it. That is a possibility with these models and heads can crash on them trying this. Had the same model as you do in here last week and the heads scored the platter and it was impossible to get off data on it. The heads crashed on it and it was acting very similar to what you are saying accourding to the person who brought this here for me.

I would like to tell you to stop now but I am not sure that you will take this advise from me. I will just warn you that it is possible to lose all your data on this drive and the heads can crash on this one. Especially if you have a lot of errors on it. It might not be possible with any software to clone this drive for you. If you continue on then the heads can get weaker and it is a possibility that anyone who tries to clone this drive for you will have data loss on it.

[Chico]

Hi guys

I appreciate your help.

Vulcan,
Yes I understand that the live cd makes those errors during boot because it tries to mount the drives. But I also understand that all of the mounting problems are irrelevant as what I'm trying to do is copy the drive with dd, without mounting it. Is this correct?
I don't want to abuse from your help, and by asking more and more it's not directed towards you in particular, just towards anyone that can help me by reading this topic. I'm sorry for going too fast haha

I have verified the backed up data I have and it seems I'm not missing anything that will make our life extremely miserable. We can live with it. We've had a couple of days already to get used to the scenario as well.
So yes, we can afford to eventually wear out the drive whilst trying to retrieve data from it. That way at least I'll make sure that it's 100% not working when I RMA it haha

But yes, I would still like to get whatever I can back.

poehere I appreciate your willing to help me but indeed we're half a world away from each other Thanks also for the contact of that guy from PCImage. I don't know if it's worth me bothering him to ask about this stuff. I don't think we can afford more than just a couple dozen euros with this (call me cheap if you want). I don't even know if he might feel insulted given that I don't have much money to spend!

I have thought of something, please tell me if it's completely nonsense. Could this be a problem with the partition table and if so, could I try formating the drive (i mean do a quick format, just rewriting the partition table) and then try to retrieve any data from it? I know by doing this I can loose everything, but it seems whatever I do I just risk loosing everything anyway. Please let me know if there is any chance this might work.
The most important information is on the 2nd partition, too.

Otherwise, Vulcan, would you be so kind to elaborate more on what you stated, when you have the time?

- attempting a reverse clone; and/or
- preventing all ddrescue retries and "splitting" for at least one pass of an attempted clone; and/or
- using raw access (bypassing the Linux kernel cache) to prevent extra reads (and retries) being done by the kernel.

What exactly is this and how can I try these things?


Anyway, it's also time to buy another drive. I would like to ask you if you recommend any particular brands over others? I've always used seagate momentus (7200 and 5400) and hitachi travelstars (7200). I usually prefer momentus over travelstars noise-wise and the 5400 momentus are usually good value for money. Do you guys have any info on more failure-prone brands than others?
250Go is about right, but I suppose I might go for the 320Go since there isn't much price difference. 500Go's a bit out of budget, plus my GF doesn't really need the extra space - she'll need it on an external drive rather, for backups!

[Vulcan]

But I also understand that all of the mounting problems are irrelevant as what I'm trying to do is copy the drive with dd, without mounting it. Is this correct?

No - for multiple reasons, you do not want the Live CD to mount the original disk. So it is not totally irrelevant - it isn't related to the fact that the original disk has a problem, but it could be making your recovery harder.

I have thought of something, please tell me if it's completely nonsense. Could this be a problem with the partition table
[...]
Please let me know if there is any chance this might work.

No chance at all. Do not attempt to write to the disk.

Otherwise, Vulcan, would you be so kind to elaborate more on what you stated, when you have the time?

- attempting a reverse clone; and/or
- preventing all ddrescue retries and "splitting" for at least one pass of an attempted clone; and/or
- using raw access (bypassing the Linux kernel cache) to prevent extra reads (and retries) being done by the kernel.

What exactly is this and how can I try these things?

These are different techniques which can be used with ddrescue - read its docs.

[Chico]

Ok great. I will RTFM.

Any chance of advising on which hard drives to buy?

[Chico]

Hi all

Just a quick update on my adventures with this drive.

I finally got around to start imaging my drive with ddrescue yesterday. But after a couple of hours, it would start displaying i/o errors. I managed to retrive around 12GB of information which is partly corrupted since windows scatters files all over the place. But I think I got a lot of important things! I'm really happy. Then I learned how to use testdisk to retrieve that data from the imaged clone.

Then, I thouht I would skip the step with ddrescue since there is no way I can image the whole drive (when it reaches the bad sectors it goes into i/o error) and access the damaged drive directly with testdisk! Learning how to use testdisk I realized that I could do this. So that's what I've been doing for the whole day now. I've managed to retrive pretty much everything I needed!

So I'm very happy with this new learning experience.